The world is now awash in data, but that doesn’t always mean we’re better informed. Without a way to analyze and draw accurate insights from this information, we’re prone to making false assumptions that lead us astray. According to Edward Tufte, a best-selling author and pioneer in data visualization, there are a few best practices organizations can follow to avoid common errors in analyzing and visualizing data. His technical insights can help your business not only track performance and industry trends more accurately, but also realize the full potential of the information your analytics team collects.
Ask these three questions to make sure your team is turning your data into visuals that will maximize the business value of their insights.
1. What are the design principles behind your data visualizations?
It’s smart to base visualization practices on proven design principles that emphasize clarity, concision, and communication. Doing so can help promote efficiency, prevent wasted work, and maximize the value of your team’s data analysis, by making sure each conclusion is presented effectively.
If your team hasn’t established a set of design principles, try looking at best-in-class examples for inspiration, including those outside the business world. A 1957 topographic map of the Swiss Alps, for instance, embodies multiple aspects of good data presentation: Colors are content driven and realistic; local details are displayed in a larger context; size of type is proportional to object labeled; and each data point has purpose. Motion and depth are also important, as they can help create multiple layers of additional meaning or provide your audience with a fresh perspective on the issue you’re attempting to solve. Ensuring your team applies similar best practices when visualizing data can go a long way toward making the most of your analytics.
2. How are you accounting for potential bias?
Because data analysis often relies on human interpretation, bias can creep in. When this happens, the data may become unreliable. One way to help your analysts account for this tendency is by asking them to follow proactive analytical strategies.
For example, this can be accomplished through data analysis that is problem-oriented rather than method-oriented. Ask your analysts to focus on the truths, expected or unexpected, they’re trying to uncover behind their dataset. You could encourage this type of analysis by having your team specify all tables, analytical methods, and models ahead of time—an approach the FDA is already using. The team could also limit its analyses to a set number of predefined models to prevent them from trying to fit the data to a predetermined conclusion. These strategies can lead to more accurate, and ultimately more useful, results.
3. How are you protecting the health of your data?
Data is vulnerable to a variety of threats, from biased analysis and misinterpretation to incorrect measurements. However, there are several strategies you can put in place to protect the health of your data—and be more certain of the insights that come out of it.
For example, because it’s unrealistic to expect any one person to detect all the contradictions and problems within a dataset, machine learning and AI tools can help your team analyze data and find errors. Such automation can lead to more precise, more efficient conclusions. Another recommendation is to have your team consistently check the accuracy of their metrics. This could be achieved by having your analysts regularly re-evaluate their methodologies to reaffirm the ways they’re measuring their numbers. Techniques such as these can help safeguard the validity of your analytics—from the moment the data is generated to when conclusions are drawn.
Tufte emphasizes that the key to good data analysis is having an open mind, but not an empty head. It’s essential to be able to embrace data and dig into it, even when it doesn’t show you want you want. Cloud-based data analysis tools like Power BI will let your team of researchers do just this, as well as build informative reports and visualizations that meet the standard for turning information into accurate conclusions.
Watch Tufte’s presentation on “The Future of Data Analysis” for more insights into how you can take full advantage of your data. To stay up to date on the latest news about Microsoft’s work in the cloud, visit https://cloudblogs.microsoft.com.